CN-122020754-A - Static satellite irradiance correction method, device, equipment and medium for distributed photovoltaic

CN122020754ACN 122020754 ACN122020754 ACN 122020754ACN-122020754-A

Abstract

The invention discloses a static satellite irradiance correction method, device, equipment and medium for distributed photovoltaic. The method comprises the steps of obtaining irradiance prediction data of a static meteorological satellite of a target area and space-time auxiliary features corresponding to the irradiance prediction data, determining clear sky irradiance references corresponding to the target area at a prediction time based on radiation transmission simulation, calculating sun geometry, static satellite observation geometry and corresponding grouping indexes based on geographic positions, the prediction time and static satellite sub-star point parameters of a distributed photovoltaic station, constructing a geometrical consistency absolute position deviation item based on the grouping indexes, inputting the irradiance prediction data, the space-time auxiliary features, the clear sky irradiance references and the geometrical consistency absolute position deviation item into a correction model, outputting correction irradiance based on the model, and carrying out physical consistency back-casting constraint on the correction irradiance to determine corrected irradiance prediction. The irradiance prediction stability and accuracy are improved.

Inventors

TANG MING
CHEN XIN
Zha Junjie
Yu Jinghang
XIA FEI
LIU CHANG
LI YAQIAO
WANG ZHIKANG
LIU ZHE
CHENG XINYUN
HE JINLING
LIU ZIHAN
WANG YUANBING

Assignees

国网江苏省电力有限公司信息通信分公司
南京信息工程大学

Dates

Publication Date: 20260512
Application Date: 20260126

Claims (10)

1. A stationary satellite irradiance correction method for distributed photovoltaics, comprising: Acquiring irradiance prediction data of a static meteorological satellite in a target area and space-time auxiliary features corresponding to the irradiance prediction data; determining a clear sky irradiance benchmark corresponding to the target area at a predicted moment based on radiation transmission simulation; Calculating sun geometry and static satellite observation geometry based on geographic position, prediction time and static satellite sub-star parameters of a distributed photovoltaic station, determining grouping indexes corresponding to the sun geometry and the static satellite observation geometry, and constructing a geometrically consistent absolute position deviation item based on the grouping indexes; Inputting the irradiance prediction data, the space-time auxiliary features, the clear sky irradiance standard and the geometric uniform absolute position deviation item into a correction model, and determining correction irradiance based on a model output result, wherein the correction model is a pre-trained hierarchical transform model based on window attention; and performing physical consistency back-projection constraint on the corrected irradiance to determine corrected irradiance prediction.
2. The method of claim 1, wherein the irradiance prediction data comprises at least one of total irradiance, direct irradiance, and component plane irradiance, wherein the spatio-temporal assist features comprise at least one of a time stamp, positioning information of the distributed photovoltaic site, cloud-related products, quality identifications, and historical error statistics, and wherein the cloud-related products comprise at least one of cloud top height, cloud phase, and cloud optical thickness.
3. The method of claim 1, wherein determining a clear sky irradiance reference for the target area at a predicted time based on radiation transmission simulation comprises: selecting a radiation transmission simulation tool, and inputting the geographic position, the prediction time and typical atmospheric state parameters of a target area, wherein the typical atmospheric state parameters comprise at least one of air pressure, aerosol content, water vapor content and ozone concentration; And simulating a propagation path of solar rays entering the earth atmosphere from the outer space to obtain the irradiation intensity reaching the ground of the target area under the cloudless condition, and taking the irradiation intensity as the clear sky irradiance standard.
4. The method of claim 1, wherein the solar geometry comprises a solar zenith angle and a solar azimuth angle, wherein the stationary satellite observation geometry comprises an observation zenith angle and an observation azimuth angle, wherein the grouping index comprises a geometry grouping index and a latitude window index, and wherein the determining a grouping index corresponding to the solar geometry and the stationary satellite observation geometry comprises: The geometrical grouping index is obtained by respectively carrying out preset interval box quantization on the differences of the solar zenith angle, the observed zenith angle, the solar azimuth angle and the observed azimuth angle and then combining the differences; and carrying out window division on the target area based on a preset latitude step length to obtain the latitude window index.
5. The method of claim 4, wherein said constructing geometrically consistent absolute positional deviation term based on said grouped index comprises: And matching a corresponding deviation submatrix from a pre-constructed deviation table based on the geometric grouping index and the latitude window index to form a deviation term matched with the current observation geometry.
6. The method of claim 1, further comprising, prior to said inputting the irradiance prediction data, the spatio-temporal assist feature, the clear sky irradiance reference, and the geometrically consistent absolute positional deviation term into a correction model: Acquiring historical static satellite irradiance prediction data, corresponding space-time auxiliary features, a historical clear sky irradiance benchmark obtained through radiation transmission simulation, and actual measurement irradiance data of a distributed photovoltaic station; constructing a training data set based on the historical stationary satellite irradiance prediction data, the corresponding space-time assist features, the historical clear sky irradiance benchmark and the measured irradiance data; Preprocessing a training data set, normalizing the historical static satellite irradiance prediction data and the clear sky irradiance standard to obtain a clear sky index, and dividing the clear sky index into a training set, a verification set and a test set according to a preset proportion; Training a pre-constructed hierarchical Transformer model based on training set iteration, dividing input features into attention windows according to window sizes in each training round, merging geometrically consistent absolute position deviation items in window self-attention calculation, and optimizing model parameters and deviation parameters in a deviation table through back propagation; And monitoring the model training effect through the verification set, stopping training when a preset training period number is reached, evaluating the model generalization capability through the test set, and selecting a model with the minimum test set error as the correction model.
7. The method of claim 1, wherein the physical consistency back-casting constraint comprises at least one of a non-negative constraint, a clear sky upper bound constraint, a night constraint and a climbing rate constraint, wherein the non-negative constraint is that the corrected irradiance prediction is not less than zero, the clear sky upper bound constraint is that the corrected irradiance prediction does not exceed a preset multiple of a clear sky irradiance reference, the night constraint is that the corrected irradiance prediction is set to zero or to a near zero range when a solar altitude is below a preset altitude threshold, and the climbing rate constraint is that the change rate of the corrected irradiance prediction at adjacent moments is limited or smoothed.
8. A stationary satellite irradiance correction device for distributed photovoltaics, comprising: the data acquisition module is used for acquiring irradiance prediction data of the static meteorological satellite in the target area and space-time auxiliary features corresponding to the irradiance prediction data; The reference determining module is used for determining a clear sky irradiance reference corresponding to the target area at the predicted moment based on radiation transmission simulation; The deviation term construction module is used for calculating sun geometry and static satellite observation geometry based on the geographic position, the prediction time and the static satellite sub-star point parameters of the distributed photovoltaic station, determining grouping indexes corresponding to the sun geometry and the static satellite observation geometry, and constructing a geometrical consistency absolute position deviation term based on the grouping indexes; The correction module is used for inputting the irradiance prediction data, the space-time auxiliary features, the clear sky irradiance standard and the geometric uniform absolute position deviation item into a correction model and determining correction irradiance based on a model output result, wherein the correction model is a pre-trained hierarchical transform model based on window attention; And the constraint module is used for carrying out physical consistency back-casting constraint on the corrected irradiance so as to determine corrected irradiance prediction.
9. An electronic device, the electronic device comprising: At least one processor; and a memory communicatively coupled to the at least one processor; Wherein the memory stores a computer program executable by the at least one processor to enable the at least one processor to perform the stationary satellite irradiance correction method for distributed photovoltaics of any of claims 1-7.
10. A computer readable storage medium storing computer instructions for causing a processor to implement the stationary satellite irradiance correction method for distributed photovoltaics of any one of claims 1-7 when executed.

Description

Static satellite irradiance correction method, device, equipment and medium for distributed photovoltaic Technical Field The invention relates to the technical field of photovoltaic power prediction and satellite remote sensing radiation processing, in particular to a static satellite irradiance correction method, device, equipment and medium for distributed photovoltaic. Background The distributed photovoltaic stations are large in number, wide in distribution and large in underlying surface difference, and are obviously influenced by rapid evolution of local cloud systems, shielding of aerosol and terrain and the like, so that the distributed photovoltaic output shows stronger spatial heterogeneity and time-varying property. Irradiance is a key driving quantity affecting photovoltaic output, and irradiance prediction quality directly determines the upper limit of distributed photovoltaic output prediction. The stationary meteorological satellite has the advantages of high time resolution, continuous coverage and the like, and is often used for short-term irradiance estimation and extrapolation. However, under different latitudes and different observation angles, systematic deviation is easy to generate in static satellite irradiance prediction, and meanwhile, non-stationary errors can be introduced by factors such as parallax, shadow propagation and the like caused by cloud system rapid generation and extinction and cloud top height. In the existing correction method, sliding average, empirical piecewise regression or simple deviation correction is mostly adopted, so that the coupling relation of 'observation geometry-error morphology' is difficult to be explicitly expressed, and the correction result is difficult to be constrained to meet the physical consistency, so that generalization and insufficient stability in a distributed scene are caused. Disclosure of Invention The invention provides a static satellite irradiance correction method, device, equipment and medium for distributed photovoltaic, which are used for solving the problems that the existing correction method is sensitive to observation geometry, lacks of physical consistency constraint and is insufficient in generalization of a distributed scene. According to an aspect of the present invention, there is provided a stationary satellite irradiance correction method for distributed photovoltaics, the method comprising: Acquiring irradiance prediction data of a static meteorological satellite in a target area and space-time auxiliary features corresponding to the irradiance prediction data; determining a clear sky irradiance benchmark corresponding to the target area at a predicted moment based on radiation transmission simulation; Calculating sun geometry and static satellite observation geometry based on geographic position, prediction time and static satellite sub-star parameters of a distributed photovoltaic station, determining grouping indexes corresponding to the sun geometry and the static satellite observation geometry, and constructing a geometrically consistent absolute position deviation item based on the grouping indexes; Inputting the irradiance prediction data, the space-time auxiliary features, the clear sky irradiance standard and the geometric uniform absolute position deviation item into a correction model, and determining correction irradiance based on a model output result, wherein the correction model is a pre-trained hierarchical transform model based on window attention; and performing physical consistency back-projection constraint on the corrected irradiance to determine corrected irradiance prediction. According to another aspect of the present invention, there is provided a stationary satellite irradiance correction device for distributed photovoltaics, the device comprising: the data acquisition module is used for acquiring irradiance prediction data of the static meteorological satellite in the target area and space-time auxiliary features corresponding to the irradiance prediction data; The reference determining module is used for determining a clear sky irradiance reference corresponding to the target area at the predicted moment based on radiation transmission simulation; The deviation term construction module is used for calculating sun geometry and static satellite observation geometry based on the geographic position, the prediction time and the static satellite sub-star point parameters of the distributed photovoltaic station, determining grouping indexes corresponding to the sun geometry and the static satellite observation geometry, and constructing a geometrical consistency absolute position deviation term based on the grouping indexes; The correction module is used for inputting the irradiance prediction data, the space-time auxiliary features, the clear sky irradiance standard and the geometric uniform absolute position deviation item into a correction model and determining correction irradiance based